Hatem Hajri’s research while affiliated with Innovation Research and Training Inc and other places

Object detection and tracking are essential for reliable decision-making in modern applications, such as self-driving cars, drones, and industry. Adverse weather can hinder object detectability and pose a threat to the reliability of these systems. As a result, there is an increasing need for efficient image denoising and restoration techniques. In this study, we investigate the use of image purification as a means of defending against weather corruptions. Specifically, we focus on the effect of snow on an object detector and the benefits of efficient desnowification. We find that the performance of a strong image purifying baseline (PreNet) is not constant across different levels of snow intensity, leading to a reduced overall performance in diverse situations. Through extensive experimentation, we demonstrate that adding a lightweight snow detector significantly improves the overall object detection performance without needing to modify the purification model. Our proposed weather-robust architecture exhibits a 40% performance improvement compared to a strong image purification baseline on the gas cylinder counting task. In addition, it leads to significant reductions of the computational power required to run the purification pipeline with a minimal added cost.

A potent class of generative models known as Diffusion Probabilistic Models (DPMs) has become prominent. A forward diffusion process adds gradually noise to data, while a model learns to gradually denoise. Sampling from pre-trained DPMs is obtained by solving differential equations (DE) defined by the learnt model, a process which has shown to be prohibitively slow. Numerous efforts on speeding-up this process have consisted on crafting powerful ODE solvers. Despite being quick, such solvers do not usually reach the optimal quality achieved by available slow SDE solvers. Our goal is to propose SDE solvers that reach optimal quality without requiring several hundreds or thousands of NFEs to achieve that goal. In this work, we propose Stochastic Exponential Derivative-free Solvers (SEEDS), improving and generalizing Exponential Integrator approaches to the stochastic case on several frameworks. After carefully analyzing the formulation of exact solutions of diffusion SDEs, we craft SEEDS to analytically compute the linear part of such solutions. Inspired by the Exponential Time-Differencing method, SEEDS uses a novel treatment of the stochastic components of solutions, enabling the analytical computation of their variance, and contains high-order terms allowing to reach optimal quality sampling $\sim3$-$5\times$ faster than previous SDE methods. We validate our approach on several image generation benchmarks, showing that SEEDS outperforms or is competitive with previous SDE solvers. Contrary to the latter, SEEDS are derivative and training free, and we fully prove strong convergence guarantees for them.

Detecting communities on graphs has received significant interest in recent literature. Current state-of-the-art approaches tackle this problem by coupling Euclidean graph embedding with community detection. Considering the success of hyperbolic representations of graph-structured data in the last years, an ongoing challenge is to set up a hyperbolic approach to the community detection problem. The present paper meets this challenge by introducing a Riemannian geometry based framework for learning communities on graphs. The proposed methodology combines graph embedding on hyperbolic spaces with Riemannian K-means or Riemannian mixture models to perform community detection. The usefulness of this framework is illustrated through several experiments on generated community graphs and real-world social networks as well as comparisons with the most powerful baselines. The code implementing hyperbolic community embedding is available online https://www.github.com/tgeral68/HyperbolicGraphAndGMM.

In this paper, we present an approach which relies on the use of random noises to generate adversarial examples of deep neural network classifiers. We argue that existing deterministic attacks, which perform by sequentially applying maximal perturbations on selected components of the input, fail at reaching accurate adversarial examples on real-world large scale datasets. By exploiting a simple Taylor expansion of the expected output probability under the noise perturbation, we introduce noise-based sparse (or L 0 ) targeted and untargeted attacks. Our proposed method, called Voting Folded Gaussian Attack (VFGA), achieves significantly better L 0 scores than state-of-the-art L 0 attacks (such as SparseFool and Sparse-RS) while being faster on both CIFAR-10 and ImageNet. Moreover, we show that VFGA is also applicable as an L ∞ attack and outperforms the state-of-the-art projected gradient attack (PGD) method.

To improve policy robustness of deep reinforcement learning agents, a line of recent works focus on producing disturbances of the environment. Existing approaches of the literature to generate meaningful disturbances of the environment are adversarial reinforcement learning methods. These methods set the problem as a two-player game between the protagonist agent, which learns to perform a task in an environment, and the adversary agent, which learns to disturb the protagonist via modifications of the considered environment. Both protagonist and adversary are trained with deep reinforcement learning algorithms. Alternatively, we propose in this paper to build on gradient-based adversarial attacks, usually used for classification tasks for instance, that we apply on the critic network of the protagonist to identify efficient disturbances of the environment. Rather than learning an attacker policy, which usually reveals as very complex and unstable, we leverage the knowledge of the critic network of the protagonist, to dynamically complexify the task at each step of the learning process. We show that our method, while being faster and lighter, leads to significantly better improvements in policy robustness than existing methods of the literature.

Certification of neural networks is an important and challenging problem that has been attracting the attention of the machine learning community since few years. In this paper, we focus on ran-domized smoothing (RS) which is considered as the state-of-the-art method to obtain certifiably robust neural networks. In particular, a new data-dependent RS technique called ANCER introduced recently can be used to certify ellipses with orthogonal axis near each input data of the neural network. In this work, we remark that ANCER is not invariant under rotation of input data and propose a new rotationally-invariant formulation of it which can certify ellipses without constraints on their axis. Our approach called Riemannian Data Dependant Randomized Smoothing (RD-DRS) relies on information geometry techniques on the manifold of covariance matrices and can certify bigger regions than ANCER based on our experiments on the MNIST dataset.

We investigate the problems and challenges of evaluating the robustness of Differential Equation-based (DE) networks against synthetic distribution shifts. We propose a novel and simple accuracy metric which can be used to evaluate intrinsic robustness and to validate dataset corruption simulators. We also propose methodology recommendations, destined for evaluating the many faces of neural DEs' robustness and for comparing them with their discrete counterparts rigorously. We then use this criteria to evaluate a cheap data augmentation technique as a reliable way for demonstrating the natural robustness of neural ODEs against simulated image corruptions across multiple datasets.

In this paper we show that neural ODE analogs of recurrent (ODE-RNN) and Long Short-Term Memory (ODE-LSTM) networks can be algorithmically embeddeded into the class of polynomial systems. This embedding preserves input-output behavior and can suitably be extended to other neural DE architectures. We then use realization theory of polynomial systems to provide necessary conditions for an input-output map to be realizable by an ODE-LSTM and sufficient conditions for minimality of such systems. These results represent the first steps towards realization theory of recurrent neural ODE architectures, which is is expected be useful for model reduction and learning algorithm analysis of recurrent neural ODEs.